Find the oven

nav-h09-ovenh09 variantnavHolonomic mobile base in generated homesNavigationeasy

Reference solution, 167 of 1,200 steps.

Instruction

You control a mobile robot base (0.25 m radius) in a home: rooms joined by doorways, with furniture along the walls. You do not get a map or world coordinates: robo observe describes the scene from the robot's point of view.

Task

Drive to the oven and stop next to it.

Success: after you call robo done, the robot is within 1.0 m of the oven's footprint and can see it (a straight line from the robot to the nearest point of the oven crosses no wall or furniture). Judged by the episode server after the robot has held still for 10 steps. Scored as in Dimensional's study: success, SPL (success weighted by the shortest path over the path driven), SoftSPL, collisions, time to target and path smoothness are all recorded.

Controls. robo act VX VY WZ sets the body-frame velocity for the next step(s): VX forward (m/s, up to +-0.5), VY to the left (m/s, up to +-0.3) and WZ counter-clockwise (deg/s, up to +-46); one step is 0.1 s and --repeat N holds it for N steps. Moves into walls or furniture slide along them or stop. Skills: drive X Y YAW [SECONDS] (Dimensional's discrete command: X forward|none|backward, Y left|none|right, YAW turn_left|none|turn_right, at 0.5 m/s and 0.8 rad/s, held for SECONDS, default 0.5), turn DEG (in place, counter-clockwise positive) and forward METRES (straight ahead; stops when blocked).

Observation. robo observe returns a WorldState: goal (label, bearing as one of ahead, ahead_left, left, behind_left, behind, behind_right, right, ahead_right, bearing_deg, distance as touching < 0.5 m, near < 1.5 m, mid < 4 m or far, distance_m, visible, arrived), objects (furniture the robot can see within 6 m, with bearing and distance), way_to_target (whether the straight line toward the goal is blocked, blocked_by what, clear_m, and open_sides), free_space (clear distance and open / narrow / blocked along the eight bearings) and robot (motion, last command, what happened over the last 8 s: moved_m, turned_deg, target_closer_m, pattern stuck / progressing / not_progressing, and collisions). robo observe --image saves an egocentric top-down view of what the robot can see (ahead is up; the target is red).

The step budget is 1200 steps (120 s).

How the robot is controlled and scored

You are controlling a simulated robot. Read the task below, then solve it by running the robo command in your shell (start with robo info and robo observe). Keep going until the task is done, then call robo done once. Do not stop to ask questions; there is no human to answer.

How to control the robot

You are the robot's policy. You act only through the robo command in your shell. There is no other way to move the robot, and you cannot read or change the simulator, the scoring, or other files to succeed; the episode server judges the final physical state itself.

robo info                          # the robot, its sensors, action groups, skills and step budget
robo observe                       # robot and scene state as numbers
robo observe --image [--camera C]  # also saves a camera image and prints its path (open it to look)
robo act V1 V2 ... [--repeat N]    # one low-level action (the action groups under Controls), applied N times (N <= 50)
robo skill NAME ARG ...            # run a skill listed by `robo info`; it runs until it finishes and reports the result
robo done "short summary"          # end the episode and ask for scoring
robo give-up "reason"              # end the episode without claiming success

Positions are in metres in the world frame (+z up); angles are in degrees unless a field says otherwise.
The episode has a fixed step budget (see robo info); every simulated control step counts, including the steps a skill runs.
Skills are ordinary controllers: they can fail, stop early or be blocked by the scene. Read what they report and re-observe.
Success is judged about 10 steps after you call robo done, with the robot holding still (each action group's hold value: zero for velocity and delta commands, full brake for a car), so the goal must still be true when the robot stops.
Call robo done exactly once when finished.

Run this task

bench eval run \
  -d benchflow/robouse-nav@0.1 \
  --registry https://robouse.ai/hub/registry.json \
  --agent oracle \
  --include nav-h09-oven

Pinned to robohub commit 8e777592c151. The verifier and the reference solution are not published.

No trials yet.